Codes Employed

The components of the end-to-end forecast system used in the RRTMG Sensitivity Test included:

    • WRF Preprocessing System (WPS) (v3.3.1)

    • WRF-ARW model (v3.3.1)

    • Unified Post Processor (UPP) (v1.0)

    • Model Evaluation Tools (MET) (v3.0.1) *Includes PB2NC error       described here

    • NCAR Command Language (NCL) for graphics generation

    • Statistical programming language, R, to compute confidence        intervals

Domain Configuration

    • Contiguous U.S. (CONUS) domain with 15-km grid spacing

Click thumbnail for larger image.

    • 403 x 302 gridpoints, for a total of 121,706 horizontal gridpoints

    • 56 vertical levels (57 sigma entries); model top at 10 hPa

    • Lambert-Conformal map projection

Sample Namelists

    • namelist.wps

    • namelist.input based on the AFWA Operational Configuration

    • namelist.input with RRTMG replacement for long-wave and        short-wave radiation

AFWA Reference Configuration

    • Sample namelist.input

RRTMG-replacement Configuration

    • Sample namelist.input

Other run-time settings

    • Long timestep = 90 s; Acoustic step = 4
    • Calls to the boundary layer and microphysics parameterization        were made every time step
    • Calls to the cumulus parameterization were made every 5        minutes
    • Calls to radiation were made every 30 minutes

The end-to-end system for this test did not include a data assimilation component.

Initial and Boundary Conditions

    • Initial conditions (ICs) and Lateral Boundary Conditions
       (LBCs): 0.5 x 0.5 degree Global Forecast System (GFS) model

    • Lower Boundary Conditions (LoBCs): Agricultural
       Meteorological Model (AGRMET) System output

    • SST Initialization: Fleet Numerical Meteorology and        Oceanography Center (FNMOC) daily, real-time sea surface        temperature (SST) product

Cases Run

• Forecast Dates: 2 June 2008 - 31 May 2009

• Initializations: Every 36 hours, including both 00 and 12 UTC

• Forecast Length: 48 hours; output files generated every 3 hours

The tables below list the forecast initializations that failed to complete the end-to-end process due to the reasons described in the table. All incomplete forecasts were due to missing or bad input data sets, not model crashes.

Verification

The Model Evaluation Tools (MET) package, comprised of:

    • grid-to-point comparisons - utilized for surface and upper air        model data
    • grid-to-grid comparisons - utilized for QPF

was used to generate objective verification statistics, including:

    • Bias-corrected Root Mean Square Error (BCRMSE) and Mean
       Error (Bias) for:
        • Surface: Temperature (2 m), Dew Point Temperature (2 m)            and Winds (10 m)
        • Upper Air: Temperature, Dew Point Temperature and Winds

    • Equitable Threat Score (ETS) and Frequency Bias (FBias) for:
        • 3-hr and 24-hr Precipiation Accumulation intervals

Each type of verification metric is accompanied by confidence intervals (CIs), at the 99% level, computed using a parametric method for the surface and upper air variables and a boostrapping method for precipitation.

Both configurations were run for the same cases allowing for a pair-wise difference methodology to be applied, as appropriate. The CIs on the pair-wise differences between statistics for the two configurations objectively determines whether the differences are statistically significant (SS).

To establish practical significance (PS) in order to examine SS pair-wise differences that are potentially more meaningful, the data was censored to only look at differences that were greater than the operational measurement uncertainty requirements as defined by the World Meteorological Organization (WMO). The following criteria were applied to determine PS pair-wise differences between the configurations and versions for select variables: i) temperature and dew point temperature differences greater than 0.1K and ii) wind speed differences greater than 0.5 ms^-1.

Area-averaged verification results were computed for the full domain, as well as 14 sub-domains.